REFRACTORIES 483 volume, w = saturated weight, s = suspended weight, in grams, and 5 = density of the liquid. A modified pycnometer has also been used by Schurecht1 for the determination of the volume of test specimens. It consists essentially of a wide-mouth cylindrical glass vessel provided with a ground in; hollow stopper which has a tube in its center. The volume of this piece is computed from the relation: v = (w — wi -f b)/5, where v = volume of specimen; w — weight of bottle plus liquid in grams; Wi = weight of bottle plus liquid plus specimen; 6 = weight of saturated test piece, and 5 = specific gravity of liquid. With increase in temperature above a minimum point the apparent volume of most refractories decreases steadily, due to colloidal changes or to the influence of fluxes which tend to bring about softening of the mass. In general, most refractory oxides undergo contraction upon being heated which in a large measure must be described as a surface-tension phenomenon. This force is active in reducing the superficial area of the mass and theoretically tends to reduce the volume to the shape of a sphere. To use a homely simile the mass contracts as if it were uniformly acted upon by a tension similar to that of rubber stretched upon a ball. At lower temperatures the rigidity of the system is too great to cause appreciable contraction but at higher heats the effect becomes more and more manifest as may be readily demonstrated by heating alumina, magnesium oxide, thorium oxide, zirconium oxide, etc., to temperatures greatly below their fusion points proper. This contraction increases rapidly in the presence of fluxes so that for instance the addition of a small amount of boric acid will cause alumina to reach constant volume at 1,750°C., about 300° below its softening or melting temperature. The presence of basic oxides in clay and siliceous materials and of acid ones in the basic refractories tends to form eutectic combinations which depress the softening temperature of the whole and at the same time bring about a marked decrease in the viscosity of the system. This condition may cause marked shrinkage at furnace working temperatures. For this reason it is desirable that the refractories be fired, if possible, to a temperature higher than that at which they are used. Porosity.—Coincident with the shrinkage in volume there occurs contraction of the pore space. The reduction in porosity may therefore be taken as a criterion of the gradual softening of the mass and being determined quite readily it is often used for the purpose of establishing the refractory character of a material. This method is not so sensitive with reference to small temperature effects as the measurement of volume but is amply accurate for most purposes. The determination of porosity involves the weighing of the fired specimen in air, weighing the piece when completely saturated with kerosene or water and finally when suspended from the beam of the balance and immersed in the liquid at a definite temperature. The first determination requires that the specimen be previously dried to remove any moisture. In bringing about complete saturation the piece is usually allowed to boil in water under a partial vacuum for 5 hr. and is then allowed to cool in water down to the normal temperature. The porosity, or pore space, expressed in per cent of the volume of the specimen, is computed from the Purdy formula: Per cent p = 100(w — d)/w — s, where, w = weight in air of specimen, saturated with water, in grams, d = weight in air of dry specimen, s = weight of specimen suspended in water, and per cent p = pore space, in per cent of the external volume. i Journ. Ceram. Soc.t Vol. 1, p. 556-58.